Has anybody given much thought to "active" health monitoring of services, where SMF (either periodically or on demand) performs some service-specific "ping" of the service to determine whether it's healthy?
The idea would be to detect services that are "running", in a process sense, but that have stopped responding. I may need to implement something like it in SMFlite to compensate for the lack of contracts (which prevents me from doing even "passive" health monitoring), and it'd be nice if whatever I do was coordinated with a long-term true-SMF plan.
